Skip to main content
Premium Trial:

Request an Annual Quote

NanoString-Based Genotype, Phenotype Test Rapidly Detects Antimicrobial Resistance


NEW YORK – A molecular method to identify bacterial infections and detect their antibiotic resistance phenotypes is generalizable to many types of bacteria and can produce results from clinical blood culture samples in about four hours, according to a new study.

The technique, called combined genotypic and phenotypic AST though RNA detection (GoPhAST-R), uses NanoString's enzyme-free hybridization-based technology, and the firm now intends to commercialize the method through an industry partnership.

Phenotypic antimicrobial susceptibility testing (AST) can take many days because it requires measurement of bacterial growth in the presence of different antibacterial drugs. Rapid AST methods incorporating MALDI-TOF, PCR-based genotypic resistance tests, or phenotypic tests like microscopy offer huge improvements in time to results over traditional methods. Unfortunately, genotypic tests do not predict resistance with total accuracy, and microscopy requires some bacterial growth, so the time to results can be limited by rates of cell division.

As described this week in Nature Medicine, the GoPhAST-R assay is able to leap these hurdles because it relies on detecting a transcriptional signature.

"In five minutes, a bug that is stressed with an antibiotic will change its expression profile," said Deborah Hung, co-director of the Infectious Disease and Microbiome Program at the Broad Institute and a senior author on the study.

Indeed, the biology of the transcriptional response is extremely fast, she said, and the maximum time the team ever needed to see a response was 30 minutes.

The GoPhAST-R method was developed by initially sequencing messenger RNA of susceptible or resistant bacteria after application of antibiotics.

For different combinations of bacterial strains and drugs, transcriptional signatures were revealed that indicated a reaction to antibiotic, and strains could be classed as resistant if they did not react.

The research team used machine learning to figure out the smallest number of genes that would distinguish susceptible versus resistant bacteria, Hung said, and applied different statistical classifiers to turn the transcriptional response into a binary call of susceptible or resistant.

"We showed that you can follow just a few genes — 10 is more than you need, but that is what we use," Hung said.

Importantly, in the study the team also demonstrated a clinical application, specifically using the method to rapidly determine ciprofloxacin susceptibility in blood culture bottles that grew gram-negative rods from the clinical microbiology laboratory at Massachusetts General Hospital. GoPhAST-R was able to distinguish three susceptible from three resistant E. coli strains, while two clinical K. pneumoniae species were found to be susceptible.

The Nature Medicine study also showed the method could be expanded to include simultaneous profiling of additional transcripts, including genetic resistance determinants like the KPC, NDM, OXA-48, IMP and VIM carbapenemase families, and extended-spectrum beta-lactamase (ESBL) gene families.

For the detection of the mRNA panels, the team initially used the quantitative fluorescent hybridization of the NanoString nCounter platform. To speed up the process further, they then ported the assay onto the next-generation Hyb & Seq platform from NanoString. This technology accelerates hybridization by using smaller unlabeled reporter probes and faster optical scanning, the study authors noted. Using Hyb & Seq, GoPhAST-R measured susceptibility signatures to the antibiotic meropenem and carbapenemase content in less than four hours.

Broad-NanoString collaboration

The foundational work underlying the technology was initially described in a 2012 PNAS paper, and Hung explained that the NanoString technology helped motivate the development. "We started on it almost when NanoString began," she said, because the lab realized the platform was well suited for this type of analysis.

Joe Beechem, NanoString's chief science officer, noted in an interview that the company has worked closely with Hung's team ever since, codeveloping much of the technology for the AST method, and that NanoString is also on the original IP.

As with the firm's Prosigna test, the firm was able to narrow an RNA signature from a large number of transcripts down to a small number appropriate for a clinical assay, he said. "From the transcriptional reprogramming that we measure globally with Illumina-based sequencers, we generated unique signatures for every one of the bug-drug combinations," Beechem said.

Last year, the team presented preliminary data on using the Hyb & Seq to perform phylogeny-informed rRNA-based strain identification, and earlier this year it published a description of Phirst-ID in Scientific Reports. That method uses NanoString platforms and 180 pairs of hybridization probes that recognize and distinguish the 16S and 23S ribosomal RNAs from 98 clinically relevant bacterial pathogens, to rapidly identify and classify bacteria.

The lab has also created a method to purify circulating cells from whole blood, as previously reported, and Hung said they have gotten good results using this technology coupled with GoPhAST-R but had not run enough experiments yet to include that data in the Nature Medicine paper. Beechem said that the microfluidic technology was licensed by NanoString, and the company has validated it and can now make the microfluidics in house.

"This is an active development program ... we have the whole workflow now inside NanoString, where we do the Phirst-ID and the GoPhAST-R, like a one-two punch," Beechem said. He added that the firm has also now integrated the whole blood pathogen concentration piece at its Seattle lab.

The Hyb & Seq chemistry enables increased sensitivity and speed needed for a clinical AST assay, compared to the firm's original nCounter barcodes, Beechem said. The R&D team is now making the workflow completely automated within the instrument subcomponents, which he thinks might halve the four-hour time.

Now, "With the Hyb & Seq chemistry we have an unlimited ability to keep making these panels larger and more universal by including more bug-drug combinations," Beechem said.

Importantly, the team was also able to show that it has a development pipeline in place and is ready for any new drug-bug combinations that might come along. Beechem suggested that the process could take less than four weeks to generate an accurate 10-target transcriptional signature of resistance.

The NanoString license is non-exclusive, so it doesn't preclude developing the method on other platforms, Hung noted. However, "NanoString is way ahead, particularly because their platform is well-designed for the multiplex detection that we need and that will allow us to capture both the phenotype and the genotype," she said.

Doug Farrell, vice president of investor relations and corporate communications at NanoString, said that the firm is currently looking at partnerships for commercialization of the technology.

"We have an extensive commercial sales channel, but it is not focused on infectious disease," Farrell said, so the firm now has an active partnering dialog with companies specializing in rapid ID and AST. NanoString is intent on bringing the system to the clinic to meet unmet needs in rapid AST, but the precise regulatory path would need to be mapped out in partnership.

Now, "Peer-reviewed publications with blue chip organizations like the Broad really underscore the value and uniqueness of the technology and help potential partners get comfortable with the opportunity ... and [see] that it has meaningful competitive and first-mover advantage," Farrell also said.